Tempting though it is to dismiss the current fad for ‘unskewed polls’ as the usual GOP nuttiness, I have a feeling this is merely setting the stage for a new post-election conspiracy theory in which Romney won and the election was stolen.
For the uninitiated, there’s a notion going around in GOP circles that Mitt Romney would be winning the election — by 11 points, no less — if every single poll out there simply lost it “liberal bias.”
Thus we need “unskewed” polls to show the true 11 point lead that Mitt Romney really has over President Obama.
So what are unskewed polls anyway, and how do they produce a result so different to the surveys by the traditional pollsters. Well here is their explanation:
The QStarNews poll works with the premise that the partisan makeup of the electorate 37.6 percent Republicans, 33.3 percent Democrats and 29.1 percent independent voters. Additionally, our model is based on the electorate including approximately 41.6 percent self-described conservatives, 32.6 percent self-described moderates and 25.8 percent self-described liberals.
Some quick background.
When a polling firm does a poll, it doesn’t just randomly call 1,000 people and publish the results. It has go through the data and make sure it’s getting a fair sampling of the entire nation among those 1,000 people. That means, among other things, making sure it has a representative sample of male and female, old people, young people and so on. Most pollsters weight their poll to correct for the sampling bias that inevitably arises from the fact that the distribution of telephone numbers does not match the distribution of likely voters.
The approach to weighting is the ‘secret sauce’ that the professional pollsters consider their competitive advantage. The proof of their secret sauce is their ability to accurately predict the actual election outcomes. Hence organizations such as Gallup that can point to many decades of predictions that turned out to be accurate are generally considered to be more trustworthy than a brand new pollster with no track record and a prediction that is completely different to all the rest.
The starting assumption for the bizarre ‘unskewed’ methodology is the fact that the ‘unskewed’ polls report lower numbers for Republican party identification than a Rasmussen Party identification survey. Here we have the Republican approach to ‘facts’ at its best. Faced with an unpalatable fact — that Romney is losing — they pick a single data point that is in apparent contradiction, take that data point as absolute and incontrovertible truth, extrapolate out from that single data point, and dismiss all contrary evidence no matter how illogical the result.
And it’s working. According to PPP, “In North Carolina, Republicans think by a 77/9 margin that the polls are all skewed for Obama.” (If Romney loses, prepare to hear the charge, the day after the election, that “voters have a liberal bias.”)
Rasmussen’s own tracking poll (9/30) puts Obama ahead of Romney by four points (49-45). (And Rasmussen is a notoriously Republican polling outfit.)
The thing is, party identification — whether you consider yourself a Democrat or a Republican — changes over time for a lot of people. Hence the reason Rasmussen found it an interesting parameter to measure. The 37.6% Republican identification Chambers uses to ‘unskew’ the polls is 2.7% higher than the same number for the month earlier. It is also the highest number in any Rasmussen study since the firm started in 2004.
Faced with a September opinion poll showing Romney several points lower than his August standing and GOP voter identification down as well, isn’t the simplest explanation that both polls moved in the same direction for the same reason: When voters start paying attention to politics, they start to conclude that they dont’ really like either Mitt Romney or the Republican party.
Not surprisingly, Chambers has come under fire from various progressive bloggers who are poking holes in his methodology. Steve Singiser at Kos has an amusing article pointing out some of the analytical blunders in Chambers approach.
Chambers does not describe his collection methodology but the poll appears to be based on a Web poll. The site makes no mention of telephone polling, and there is a Web survey form that anyone can fill in. Self-selecting surveys are notoriously inaccurate and easy to game. They are useful as a tool of last resort, but anyone presenting one as more accurate than a random sampled telephone survey is simply deluded.
Of course, Republicans aren’t terribly motivated by accuracy.
When faced with an impartial media that was calling them out on their lies and malfeasance in the 1990s, the Republicans simply created their own media, aka Fox News. And now, when presented with polls showing their candidates going down in flames in November, the GOP will simply create its own polls.
But there’s a rub. First, polls aren’t votes. And second, convincing GOP voters that Romney is 11 points ahead risks sending the message that the game is over, Romney’s won, and there’s no reason to vote at all.
Then the GOP goes from unskewed to skewered.
[Note: Updated to edit for clarity]